| The optimization of two modular type multi-station, high-speed automated placement machines operating at Motorola was addressed. An optimization method employing improvements on classical tabu search (TS) techniques was developed, and tested in the actual machine environment.; First, a sufficiently detailed model of the machines was created using the structural specifications supplied by the machine vendor. Based on the observations gained from the detailed study of the machines and the experimentation on the model, the complete optimization scheme of the machines was elaborated. A special, “dual value” search space was established in order to cope with the high level of symmetry and modularity caused by the nature of the problem. Its integration in the tabu search framework resulted in new tabu strategies, efficiently utilizing the additional information provided by the particular characteristics of the search space.; The experiments conducted on the assembly of a variety of products at Motorola verified the advantages of the new algorithms in problem classes where the unusual structure of the search space imposes difficulties on finding satisfactory solutions.; As part of the research work, an optimization software toolkit was also developed, which has been ported under several computer platforms. |
